> From: Fajar A. Nugraha [mailto:list@fajar.net]
> Subject: [Xen-devel] Status of xenballoond?
>
> Is xenballoond still maintained?
>
> http://lxr.xensource.com/lxr/source/tools/xenballoon/xenballoond still
> use "echo $tgtbytes > /proc/xen/balloon", while
/proc/xen/balloon does
> not exists in pv_ops kernel. There''s
> /sys/devices/system/xen_memory/xen_memory0/target_kb , which seems to
> work as expected, so it might be a simple matter of adding a test to
> choose which file to write memory target to.
Hi Fajar --
The short answer is I''m not maintaining it, but patches are welcome!
xenballoond is really sample code, an in-tree proof of concept
to drive self-ballooning or "directed ballooning". See:
http://wiki.xensource.com/xenwiki/Open_Topics_For_Discussion?action=AttachFile&do=get&target=Memory+Overcommit.pdf
and
http://www.slideshare.net/xen_com_mgr/xs-boston-2008-memory-overcommit
This was only a first step (now 2-1/2 years ago), though some may
find it sufficient for their needs.
This work led me to believe that guest changes are necessary
to efficiently vary memory size, which led to Transcendent
Memory, including in-kernel self-ballooning.
http://oss.oracle.com/projects/tmem
Sadly, it is taking years to get the necessary changes into
upstream Linux, but I''m still working on that!
> Also, when writing a number lower than initial value to
> /sys/devices/system/xen_memory/xen_memory0/target_kb, "xm list"
on
> dom0 still shows old memory value, but "xm info" correctly shows
some
> memory freed. This is different with "xm mem-set" from dom0,
which
> correctly updates both values.
This is a known bug in "xm". "xl list" shows the correct
value too.
Since xm is obsolescent (and since it''s possible some product
scripts may depend on the old behavior), this has never gotten
fixed.
Dan
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel